A comprehensive analysis of prognosis prediction models based on pathway‑level, gene‑level and clinical information for glioblastoma

A potential prognostic prediction model for metastatic osteosarcoma based on bioinformatics analysis

Osteosarcoma (OS) is a malignant primary bone tumor with a high incidence. This study aims to construct a prognostic prediction model by screening the prognostic mRNA of metastatic OS. Data on four eligible expression profiles from the National Center for Biotechnology Information Gene Expression Omnibus repository were obtained based on inclusion criteria and defined as the training set or the validation set. The differentially expressed genres (DEGs) between meta- static and non-metastatic OS samples in the training set were first identified, and DEGs related to prognosis were screened by univariate Cox regression analysis. In total, 107 DEGs related to the prognosis of metastatic OS were identified. Then, 46 DEGs were isolated as the optimized prognostic gene signature, and a metastatic-OS discriminating classifier was constructed, which had a high accuracy in distinguishing metastatic from non-metastatic OS samples. Furthermore, four optimized prognostic gene signatures (ALOX5AP, COL21A1, HLA-DQB1, and LDHB) were further screened, and the prognostic prediction model for metastatic OS was constructed. This model possesses a relatively satisfying prediction ability both in the training set and validation set. The prognostic prediction model that was constructed based on the four prognostic mRNA signatures has a high predictive ability for the prognosis of metastatic OS.

Comprehensive analysis of the differential expression and prognostic value of COL1A2 in colon adenocarcinoma

Background: Colon adenocarcinoma (COAD) is a highly heterogeneous disease, which is the second most common cancer in females and third in males. Collagen type I alpha 2 (COL1A2) has been documented to be involved in the carcinogenesis of multiple tumors; however, the expression and prognostic significance of COL1A2 and its underlying mechanism in COAD remains unclarified.

Materials and Methods: The general profile of COL1A2, its expression pattern, and prognostic value were systematically assessed through various bioinformatics tools. The protein level of COL1A2 was verified in COAD patients using immunohistochemistry analysis. In addition, enrichment analyses were performed to explore the possible regulatory pathways of COL1A2 in COAD.

Results: The mRNA and protein levels of COL1A2 were significantly increased in COAD than that in normal tissues (P < 0.05). The COL1A2 expression tended to increase along with cancer stages and nodal metastasis status in COAD, while the promoter methylation levels of COL1A2 might negatively related to its mRNA expression. Survival analysis showed that COL1A2 was a reliable predictor for distinguishing the status of disease-specific survival (DSS), overall survival (OS), and progression-free survival (PFS), and might serve as a robust independent prognostic biomarker for DSS and OS in COAD patients (P < 0.05). The enrichment analysis showed focal adhesion as the most possible regulatory pathway by COL1A2.

Conclusion: Collectively, COL1A2 functioned as an independent prognostic biomarker and might be a potential therapeutic target in COAD.

FN1 overexpression is correlated with unfavorable prognosis and immune infiltrates in breast cancer

Objective: To investigate the correlation of fibronectin 1 (FN1) expression with prognosis and tumor-infiltrating immune cells in breast cancer (BRCA).

Methods: FN1 mRNA and protein expressions were analyzed through Tumor Immune Estimation Resource (TIMER), Gene Set Cancer Analysis (GSCA), Human Protein Atlas (HPA) databases, and immunohistochemical analysis. The clinicopathological characteristics and genetic factors affecting the FN1 mRNA expression were assessed by various public databases. Then, we analyzed the prognostic value of FN1 in BRCA by Kaplan-Meier plotter, receiver operating characteristic, and Cox regression analyses. Further, the UCSC Xena database was used to retrieve TCGA-BRCA expression profiles for functional enrichment analysis and immune cell infiltration analysis. The potential drugs for the BRCA patients with high- FN1 expression were identified using the connectivity map analysis.

Results: FN1 was upregulated in BRCA tissues compared with normal tissues. High FN1 mRNA expression was correlated with poor clinical outcomes and had good performance in predicting the survival status of BRCA patients. Further, Cox regression analysis showed that FN1 was an independent prognostic factor for predicting the overall survival of patients with BRCA. Moreover, hypermethylation of FN1 contributed to a better prognosis for BRCA patients. Functional enrichment analyses revealed the ECM-receptor interaction pathway and focal adhesion as the common pathways. Moreover, FN1 showed a significant association with tumor-infiltrating immune cells and immune checkpoint inhibitors. Several drugs such as telmisartan, malotilate, and seocalcitol may have therapeutic effects in BRCA patients with high FN1 expression.

Conclusion: FN1 might serve as a novel prognostic biomarker and a novel therapeutic target for BRCA. Besides, the association of FN1 with immune cells and immune checkpoint inhibitors may provide assistance for BRCA treatment.

Construction of an immune-related gene signature for the prognosis and diagnosis of glioblastoma multiforme

Background

Increasing evidence has suggested that inflammation is related to tumorigenesis and tumor progression. However, the roles of immune-related genes in the occurrence, development, and prognosis of glioblastoma multiforme (GBM) remain to be studied.

Methods

The GBM-related RNA sequencing (RNA-seq), survival, and clinical data were acquired from The Cancer Genome Atlas (TCGA), Genotype-Tissue Expression (GTEx), Chinese Glioma Genome Atlas (CGGA), and Gene Expression Omnibus (GEO) databases. Immune-related genes were obtained from the Molecular Signatures Database (MSigDB). Differently expressed immune-related genes (DE-IRGs) between GBM and normal samples were identified. Prognostic genes associated with GBM were selected by Kaplan–Meier survival analysis, Least Absolute Shrinkage and Selection Operator (LASSO)-penalized Cox regression analysis, and multivariate Cox analysis. An immune-related gene signature was developed and validated in TCGA and CGGA databases separately. The Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were performed to explore biological functions of the signature. The correlation between immune cell infiltration and the signature was analyzed by single-sample gene set enrichment analysis (ssGSEA), and the diagnostic value was investigated. The gene set enrichment analysis (GSEA) was performed to explore the potential function of the signature genes in GBM, and the protein–protein interaction (PPI) network was constructed.

Results

Three DE-IRGs [Pentraxin 3 (PTX3), TNFSF9, and bone morphogenetic protein 2 (BMP2)] were used to construct an immune-related gene signature. Receiver operating characteristic (ROC) curves and Cox analyses confirmed that the 3-gene-based prognostic signature was a good independent prognostic factor for GBM patients. We found that the signature was mainly involved in immune-related biological processes and pathways, and multiple immune cells were disordered between the high- and low-risk groups. GSEA suggested that PTX3 and TNFSF9 were mainly correlated with interleukin (IL)-17 signaling pathway, nuclear factor kappa B (NF-κB) signaling pathway, tumor necrosis factor (TNF) signaling pathway, and Toll-like receptor signaling pathway, and the PPI network indicated that they could interact directly or indirectly with inflammatory pathway proteins. Quantitative real-time PCR (qRT-PCR) indicated that the three genes were significantly different between target tissues.

Conclusion

The signature with three immune-related genes might be an independent prognostic factor for GBM patients and could be associated with the immune cell infiltration of GBM patients.

A comprehensive bioinformatics analysis to identify a candidate prognostic biomarker for ovarian cancer

BACKGROUND

This study aimed to investigate prognostic genes in ovarian cancer (OC) and to explore their potential underlying biological mechanisms through a comprehensive bioinformatics analysis.

METHODS

Common differentially expressed genes (DEGs) in 3 OC datasets from the Gene Expression Omnibus (GEO) (GSE26712, GSE18520, and GSE14407) were screened out. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed by Metascape. The protein-protein interaction (PPI) network of the DEGs was constructed using the STRING database. The prognostic value of DEGs were determined using the Kaplan-Meier plotter. The ONCOMINE and Human Protein Atlas databases were used to verify the expression levels of prognostic genes in OC. Genomic analysis of prognostic genes were also investigated by cBio Cancer Genomics Portal (cBioPortal) database, UCSC Xena browser and UALCAN. Gene set enrichment analysis (GSEA) was used to predict the possible pathways and biological processes of the prognostic genes.

RESULTS

Integration of the 3 datasets have found 879 common DEGs. A high expression of structural maintenance of chromosomes protein 4 (SMC4) was revealed in the Kaplan-Meier plotter analysis to be meaningful for the prognosis of OC and was verified at both the mRNA and protein levels. The results from cBioPortal showed that SMC4 alterations accounted for 7 to 18% of genetic alterations in OC, and the majority alterations were copy number amplifications. Finally, the GSEA results showed that samples with SMC4 overexpression were mainly enriched in the cell cycle, spliceosome, ubiquitin mediated proteolysis, and adherens junctions.

CONCLUSIONS

High SMC4 expression is linked with a poor prognosis in patients with OC and might serve as a prognostic biomarker for the disease.

Identification of C3 and FN1 as potential biomarkers associated with progression and prognosis for clear cell renal cell carcinoma

Background

Clear cell renal cell carcinoma (ccRCC) is one of the most lethal urological malignancies, but the pathogenesis and prognosis of ccRCC remain obscure, which need to be better understand.

Methods

Differentially expressed genes were identified and function enrichment analyses were performed using three publicly available ccRCC gene expression profiles downloaded from the Gene Expression Omnibus database. The protein-protein interaction and the competing endogenous RNA (ceRNA) networks were visualized by Cytoscape. Multivariate Cox analysis was used to predict an optimal risk mode, and the survival analysis was performed with the Kaplan-Meier curve and log-rank test. Protein expression data were downloaded from Clinical Proteomic Tumor Analysis Consortium database and Human Protein Atlas database, and the clinical information as well as the corresponding lncRNA and miRNA expression data were obtained via The Cancer Genome Atlas database. The co-expressed genes and potential function of candidate genes were explored using data exacted from the Cancer Cell Line Encyclopedia database.

Results

Of the 1044 differentially expressed genes shared across the three datasets, 461 were upregulated, and 583 were downregulated, which significantly enriched in multiple immunoregulatory-related biological process and tumor-associated pathways, such as HIF-1, PI3K-AKT, P53 and Rap1 signaling pathways. In the most significant module, 36 hub genes were identified and were predominantly enriched in inflammatory response and immune and biotic stimulus pathways. Survival analysis and validation of the hub genes at the mRNA and protein expression levels suggested that these genes, particularly complement component 3 (C3) and fibronectin 1 (FN1), were primarily responsible for ccRCC tumorigenesis and progression. Increased expression of C3 or FN1 was also associated with advanced clinical stage, high pathological grade, and poor survival in patients with ccRCC. Univariate and multivariate Cox regression analysis qualified the expression levels of the two genes as candidate biomarkers for predicting poor survival. FN1 was potentially regulated by miR-429, miR-216b and miR-217, and constructed a bridge to C3 and C3AR1 in the ceRNA network, indicating a critical position of FN1.

Conclusions

The biomarkers C3 and FN1 could provide theoretical support for the development of a novel prognostic tool to advance ccRCC diagnosis and targeted therapy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-021-08818-0.

PAMOGK: a pathway graph kernel-based multiomics approach for patient clustering

MOTIVATION

Accurate classification of patients into molecular subgroups is critical for the development of effective therapeutics and for deciphering what drives these subgroups to cancer. The availability of multiomics data catalogs for large cohorts of cancer patients provides multiple views into the molecular biology of the tumors with unprecedented resolution.

RESULTS

We develop Pathway-based MultiOmic Graph Kernel clustering (PAMOGK) that integrates multiomics patient data with existing biological knowledge on pathways. We develop a novel graph kernel that evaluates patient similarities based on a single molecular alteration type in the context of a pathway. To corroborate multiple views of patients evaluated by hundreds of pathways and molecular alteration combinations, we use multiview kernel clustering. Applying PAMOGK to kidney renal clear cell carcinoma (KIRC) patients results in four clusters with significantly different survival times (P-value =1.24e-11). When we compare PAMOGK to eight other state-of-the-art multiomics clustering methods, PAMOGK consistently outperforms these in terms of its ability to partition KIRC patients into groups with different survival distributions. The discovered patient subgroups also differ with respect to other clinical parameters such as tumor stage and grade, and primary tumor and metastasis tumor spreads. The pathways identified as important are highly relevant to KIRC.

AVAILABILITY AND IMPLEMENTATION

github.com/tastanlab/pamogk.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Identification of a prognostic gene signature of colon cancer using integrated bioinformatics analysis

Background

Colon cancer is a worldwide leading cause of cancer-related mortality, and the prognosis of colon cancer is still needed to be improved. This study aimed to construct a prognostic model for predicting the prognosis of colon cancer.

Methods

The gene expression profile data of colon cancer were obtained from the TCGA, GSE44861, and GSE44076 datasets. The WGCNA module genes and common differentially expressed genes (DEGs) were used to screen out the prognosis-associated DEGs, which were used to construct a prognostic model. The performance of the prognostic model was assessed and validated in the TCGA training and microarray validation sets (GSE38832 and GSE17538). At last, the model and prognosis-associated clinical factors were used for the construction of the nomogram.

Results

Five colon cancer-related WGCNA modules (including 1160 genes) and 1153 DEGs between tumor and normal tissues were identified, inclusive of 556 overlapping DEGs. Stepwise Cox regression analyses identified there were 14 prognosis-associated DEGs, of which 12 DEGs were included in the optimized prognostic gene signature. This prognostic model presented a high forecast ability for the prognosis of colon cancer both in the TCGA training dataset and the validation datasets (GSE38832 and GSE17538; AUC > 0.8). In addition, patients’ age, T classification, recurrence status, and prognostic risk score were associated with the prognosis of TCGA patients with colon cancer. The nomogram was constructed using the above factors, and the predictive 3- and 5-year survival probabilities had high compliance with the actual survival proportions.

Conclusions

The 12-gene signature prognostic model had a high predictive ability for the prognosis of colon cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12957-020-02116-y.

Survival prediction of glioblastoma patients-are we there yet? A systematic review of prognostic modeling for glioblastoma and its clinical potential

Glioblastoma is associated with a poor prognosis. Even though survival statistics are well-described at the population level, it remains challenging to predict the prognosis of an individual patient despite the increasing number of prognostic models. The aim of this study is to systematically review the literature on prognostic modeling in glioblastoma patients. A systematic literature search was performed to identify all relevant studies that developed a prognostic model for predicting overall survival in glioblastoma patients following the PRISMA guidelines. Participants, type of input, algorithm type, validation, and testing procedures were reviewed per prognostic model. Among 595 citations, 27 studies were included for qualitative review. The included studies developed and evaluated a total of 59 models, of which only seven were externally validated in a different patient cohort. The predictive performance among these studies varied widely according to the AUC (0.58-0.98), accuracy (0.69-0.98), and C-index (0.66-0.70). Three studies deployed their model as an online prediction tool, all of which were based on a statistical algorithm. The increasing performance of survival prediction models will aid personalized clinical decision-making in glioblastoma patients. The scientific realm is gravitating towards the use of machine learning models developed on high-dimensional data, often with promising results. However, none of these models has been implemented into clinical care. To facilitate the clinical implementation of high-performing survival prediction models, future efforts should focus on harmonizing data acquisition methods, improving model interpretability, and externally validating these models in multicentered, prospective fashion.

Molecular subtyping of glioblastoma based on immune-related genes for prognosis

Glioblastoma (GBM) is associated with an increasing mortality and morbidity and is considered as an aggressive brain tumor. Recently, extensive studies have been carried out to examine the molecular biology of GBM, and the progression of GBM has been suggested to be correlated with the tumor immunophenotype in a variety of studies. Samples in the current study were extracted from the ImmPort and TCGA databases to identify immune-related genes affecting GBM prognosis. A total of 92 immune-related genes displaying a significant correlation with prognosis were mined, and a shrinkage estimate was conducted on them. Among them, the 14 most representative genes showed a marked correlation with patient prognosis, and LASSO and stepwise regression analysis was carried out to further identify the genes for the construction of a predictive GBM prognosis model. Then, samples in training and test cohorts were incorporated into the model and divided to evaluate the efficiency, stability, and accuracy of the model to predict and classify the prognosis of patients and to identify the relevant immune features according to the median value of RiskScore (namely, Risk-H and Risk-L). In addition, the constructed model was able to instruct clinicians in diagnosis and prognosis prediction for various immunophenotypes.

A potential prognostic model based on miRNA expression profile in The Cancer Genome Atlas for bladder cancer patients

Background

This study aimed to construct prognostic model by screening prognostic miRNA signature of bladder cancer.

Methods

The miRNA expression profile data of bladder cancer (BC) in The Cancer Genome Atlas (TCGA) were obtained and randomly divided into the training set and the validation set. Differentially expressed miRNAs (DEMs) between BC and normal control samples in the training set were firstly identified, and DEMs related to prognosis were screened by Cox Regression analysis. Then, the MiR Score system was constructed using X-Tile based cutoff points and verified in the validation set. The prognostic clinical factors are selected out by univariate and multivariate Cox Regression analysis. Finally, the mRNAs related to prognosis were screened and the biological pathway analysis was carried out.

Results

We identified the 7-miRNA signature was significantly associated with the patient’s Overall Survival (OS). A prognostic model was constructed based on the prognostic 7-miRNA signature, and possessed a relative satisfying predicted ability both in the training set and validation set. In addition, univariate and multivariate Cox Regression analysis showed that age, lymphovascular invasion and MiR Score were considered as independent prognostic factors in BC patients. Furthermore, based on MiR Score prognostic model, several differentially expressed genes (DEGs), such as WISP3 and UNC5C, as well as their related biological pathway(s), including cell–cell adhesion and neuroactive ligand-receptor interaction, were considered to be related to BC prognosis.

Conclusion

The prognostic model which was constructed based on the prognostic 7-miRNA signature presented a high predictive ability for BC.

Comparison of pathway and gene-level models for cancer prognosis prediction

BACKGROUND

Cancer prognosis prediction is valuable for patients and clinicians because it allows them to appropriately manage care. A promising direction for improving the performance and interpretation of expression-based predictive models involves the aggregation of gene-level data into biological pathways. While many studies have used pathway-level predictors for cancer survival analysis, a comprehensive comparison of pathway-level and gene-level prognostic models has not been performed. To address this gap, we characterized the performance of penalized Cox proportional hazard models built using either pathway- or gene-level predictors for the cancers profiled in The Cancer Genome Atlas (TCGA) and pathways from the Molecular Signatures Database (MSigDB).

RESULTS

When analyzing TCGA data, we found that pathway-level models are more parsimonious, more robust, more computationally efficient and easier to interpret than gene-level models with similar predictive performance. For example, both pathway-level and gene-level models have an average Cox concordance index of ~ 0.85 for the TCGA glioma cohort, however, the gene-level model has twice as many predictors on average, the predictor composition is less stable across cross-validation folds and estimation takes 40 times as long as compared to the pathway-level model. When the complex correlation structure of the data is broken by permutation, the pathway-level model has greater predictive performance while still retaining superior interpretative power, robustness, parsimony and computational efficiency relative to the gene-level models. For example, the average concordance index of the pathway-level model increases to 0.88 while the gene-level model falls to 0.56 for the TCGA glioma cohort using survival times simulated from uncorrelated gene expression data.

CONCLUSION

The results of this study show that when the correlations among gene expression values are low, pathway-level analyses can yield better predictive performance, greater interpretative power, more robust models and less computational cost relative to a gene-level model. When correlations among genes are high, a pathway-level analysis provides equivalent predictive power compared to a gene-level analysis while retaining the advantages of interpretability, robustness and computational efficiency.

A Potential Prognostic Gene Signature for Predicting Survival for Glioblastoma Patients

OBJECTIVE

This study aimed to screen prognostic gene signature of glioblastoma (GBM) to construct prognostic model.

METHODS

Based on the GBM information in the Cancer Genome Atlas (TCGA, training set), prognostic genes (Set X) were screened by Cox regression. Then, the optimized prognostic gene signature (Set Y) was further screened by the Cox-Proportional Hazards (Cox-PH). Next, two prognostic models were constructed: model A was based on the Set Y; model B was based on part of the Set X. The samples were divided into low- and high-risk groups according to the median prognosis index (PI). GBM datasets in Gene Expression Ominous (GEO, GSE13041) and Chinese Glioma Genome Atlas (CGGA) were used as the testing datasets to confirm the prognostic models constructed based on TCGA.

RESULTS

We identified that the prognostic 14-gene signature was significantly associated with the overall survival (OS) in the TCGA. In model A, patients in high- and low-risk groups showed the significantly different OS (P = 7.47 × 10-9, area under curve (AUC) 0.995) and the prognostic ability were also confirmed in testing sets (P=0.0098 and 0.037). The model B in training set was significant but failed in testing sets.

CONCLUSION

The prognostic model which was constructed based on the prognostic 14-gene signature presented a high predictive ability for GBM. The 14-gene signature may have clinical implications in the subclassification of GBM.